國立臺灣師範大學教育心理與輔導學系 教育心理學報，民 91 ' 33 卷， 2 期， 103-124 頁
The Relationships Between Domain-Specific
Self-Concepts and Global Self-Esteem
Among Adolescents in TaiwanYu-WEI CHU
Department of Business Management Chung Kuo Institute of Technology
The primary purpose of this study is to investigate the reJationships between adolescents' global self-esteem and domain-specific seJf-concepts in Taiwan. Using Chinese Rosenberg SeJf-Esteem Scale (CRSES) and Chinese SeJf-Description Questionnaire-II (CSDQ-II), 591 participants from the junior and senior high schooJs in Taiwan reported their perceptions of gJobaJ se怔-esteemand seJf-concept. Correlation,
multipJe regression 叩alysisand analysis of variance (ANOVA) were applied to 血edata Them吋 orfi!ldings wer，巴的 follows: (a) the correlation coefficients among the factors of CSDQ-II and the CRSES were statistically significant; (b) 6 specific domains of seJf-concept (i.e., math/school, verbal, physical appe訂ance ，emotion, parent relations and peer relations) made a significant contribution to predict adolescents' global seJf-esteem; (c) gender and age differences had an impact on adolescents' perception to domain-specific self-conc句的;(d) a significant two-way interaction between grade and gender was found on the factor of peer relationshi阱，
KEY WORDS : domain-specific self四 concepts ，gender, grade, self-description
The constructs of self-concept and self-esteem have been a topic of great interest in the field
of psychology and educational research, and they have been shown to have a pervasive impact on
human behavior (Wyl時， 1979). A posítive self-concept or an adequate self-esteem is centraI to an
individual's mentaI health and adaptive functioning (Chan & Lee, 1993). General, the assessment
of adolescent self-esteem or self-concept has been regarded as an important 祖d necessary step to
lO4 . 教育，心理學報
Although the definitions of the terms, self-esteem and self-concept, could be used interchangeably by some researchers who have conducted studies in Westem culture (Pelham & Swann, 1989; Shavelson, Hubn缸， & Stanton, 1976), Watkins and Dhawan (1989) noted that self-concept and self-esteem couldn't be used interchangeably in non Westem societies such as China, Hong Kong, and Taiwan. They defined self-concept as someone's perception about hirnlherself, without involving judgement of worth, a definition which was similar to Hoelter's (1985). The te口n ， self-esteem, is referred to as self-respect in Chinese, which pertains more to self-evaluation than self-description. The distinguished definitions between self-concept and self-esteem were
adopted in this study.
Historically, self-concept has been conceptualiz活das a unidimensional construct or a global measure in the e訂ly definitions (Coopersmith、 1967; James, 1890; Rosenberg, 1965). However, more recently, self-concept has been recognized as a multidimensional construct, which consists of various specific dimensions of the self that reflect an individual's multiple roles and experiences (Marsh, 1990a). In general, previous research studies on self-concept among Chinese adolescents have been restricted to the use of a global measure (Chan & Le巴， 1993; Huang, 1971) 組d to a focus on the correlation of self-esteem and self-esteem development (Chan & L間， 1993; Lau & Cheung, 1987). Over time, the importance of assessing the multidimensionality of self-concept has been recognized; however, few studies have been designed to specifically address the relationships between domain-specific self-concepts and global self-worth among Chinese adol巴scents.
Most of the research that has been focused on self-concept was carried out in Westem countn郎，particularly in North America and Australia. Watkins and Dhawan (1989) pointed out that people from different cultures tend to think about themselves in specific ways. For example, people from Westem cultures tend to have an individualistic self-concept, with the emphasis on individual characteristics and achievement. Howev巴r， those from non- Westem cultures tend to report a collectivistic self-concept, in which the person does not think about hirnlherself so much as an individual, but rather in terms of relationships with other people (e.g., as a daughter or husband; Ch帥， 1997). Hence, the findings in Westem culture about self-esteem may not be
generalized to non-Westem cultures.
In contrast to Westem culture, Chinese socialization is characterized as a social orientation, and the Chinese people's behavior in social situations is guided by the social norm and the internalized representation of external expectation (Yang, 1981). Undoubtedly、 the Chinese practice of socia1ization and their socia1 obligations influence the youth's self-concept, because a person's self-concept is intimately tied to social confonnity, social expectation, and an attempt to obtain rewards or to gain access to desired resources (Hamid & Cheng, 1995). Therefore, it is obvious that Chinese adolescents' self-concepts are usually associated with satisfact
臺灣青少年多重向度自我概念與整體自尊關係之比較研究 105 .
Purpose of the Study
The purpose of this study was to assess the relationships between domain-specific self-concepts and global self-esteem by use of Rosenberg's Self-Esteem Scale (RSES; 1965) and the Self-Description Questionnaire-II (SDQ-II; Marsh, 1990b) with junior and senior high school students in Taiwan. Also, this researcher examined the gender and grade differences in the proportion of variance in self-esteem accounted for by different domain-specific self-concepts.
The results of 出is study were to inform teachers, parents, and school counselors to become aware of the relationships between specific self-concepts and globa1 self-esteem in adolescents. Also, this information might help teachers and parents develop interventions and programs for those adolescents with self-esteem problems.
Explored in this study were the relationships between adolescents' domain-specific self-concepts and global self-esteem in Taiwan and the differences based on di釘'erent grade and age groups. The research questions which guided this study were:
1. What are the relationships between domain-specific self-concepts and global self-esteem among adolescents in Taiwan?
2. Are there gender differences in domain-specific self-concepts among adolescents in Taiwan?
3. Are there grade differences in domain-specific selιconcepts among adolescents in Taiwan?
Review of the Literature
This review of literature was conducted to understand the definitions of esteem and self-concept, to clarify approaches and related research of unidimensiona1 and multidimensional self-concept or self司esteem，and to identify the factors that influence multidimensional self也conceptand global self-esteem.
Definitions of Self-Esteem
James (1890) expanded on his theory of the self and provided one of the earliest definitions of self-esteem. He defined self-esteem as a unidimensional construct 出atwas related to how a person felt about himJherself Cooley (1902), a sociologist, provided the notion of the looking-glas可 self. that 時， self-esteem is dependent upon one 's perception of what significant others thought of himJher. Also, Mead (1934), another sociologi泣， stressed the importance of others in relation to
106 . 教育心理學報
self and maintained that one could try to understand th巴 selffrom social interactions.
The next m句 oradvance in the definition of self-esteem was provided by Rosenberg (1965), who based his definition of self-esteem on James' concept. According to Mruk (1995), Rosenberg made three contributions to the definition of self-esteem: (a) self-esteem includes both an affective and a cognitive component; (的 self-esteem has an evaluative component; and (c) self-esteem is not only personal and psychological, but also includes social interaction. It was Rosenberg's position that self-esteem involved not only feelings but, also, cognitive conceptualizations and percept!Ons
Roser伽立s (峙的)view of self-esteem,
rsnút由h( 附7η) d血倒叫
d it 叫lOWone made an evaluation about himlherself. He added the term, worth, and associated self-esteem with both worth and evaluation which are personal and powerful aspects of the construct. Coopersrnith was the first to develop an instrument designed to measure the construct of self.司esteem， the Self-Esteem Inventory (SEI). The development of the SEI made self-est巴em a more objective and measurable construct.
In g巴neral， self-esteem has been defined as self-evaluation, and researchers have viewed it as an aff.ective part of the self, that is, the individual's regard about hislher own worth (Brandon, 1969; Coopersrnith, 1967; Daly & Diesel, 1992; Rogers, 1951). Byme (1996) noted that terms used interchangeably with self-esteem were self-regard, self-reverence, self-acceptance, self-respect, self-worth, self-feeling, and self-evaluation.
Definitions of Self二 Concept
In general, self-concept has been defined as self-desc呻tion without involving judgment of worth. It was viewed as a cognitive part of the self by Hoelter (1985) and Watkins and Dhawan (1989). Rosenberg (1979) regarded self-concept as not a real self, but rather the picture of the self. He defined self-concept as “the totality of the individual's thoughts and feelings having reference to him[/her)self as object" (p. 7). In Rosenberg's definition, self-concept not only included cognitive components but a叮叮tive ones, also. Sinúlarly, Pelham and Swann (1989) reported that self-concept had relatively independent cognitive and affective components which supported Rosenberg's definition.
Self-concept was broadly defined by Shavelson, et al. (1976) as a person's self perceptions formed through one's experience and interpretations with hislher environment. The individual is influenced especially by evaluations from significant others, reinforcements, and attributions for one's own behavior. Their definition provided a clear and concrete description of sclf-concept and indicated that 可elf-conceptincluded not only a self-description component but also self-evaluation. Byrne (1996) pointed out that terms used interchangeably with self-concept were self, self-estimation, self-identity, self司 image， self-perception, self-consciousness, self-image旬， and self-awareness.
The Distinction Between Self-Esteem and Self-Concept
Self-concept and self-esteem have been the subjects of research in the field of education and psychology over the past several decades. Many researchers have claimed that there was no
臺灣青少年多重向度自我概念與整體自尊閱(系之比較研究 107 .
difference between self-esteem and self-concept (Pelham & Swann, 1989; Shavelson et 泣， 1976) As noted by Shavelson et al . , “The distinction between self四description and self-evaluation has not been clarified either conceptually or empirically. Accordingly, the terms concept and self-esteem have been used interchangeably in the literature" (p. 414). Also, Marsh, Byme, and Shavelson (1988) suggested that self-esteem and self-concept could be used interchangeably based on the lack of empirical and theoretical evidence to support the differences between these two terms.
According to Grant (1998), some researchers have made a distinction between self-esteem and self-concept, but most of them view self-esteem as included under self-concept. Elliott (1986) identified self-esteem as the dimension of self-concept r巴latedto self-evaluation. Iheanacho (1988) pointed out that, although both self-concept and self-esteem were related self-concepts, they differed, because self-esteem was the affective p訂tof the self, and self-concepts was the cogniu-Ie
portion of the self. Watkins and Dhawan (1989) found that differences existed within self-concept and self司esteem in non-Westem societies. They supported the idea that there was a distinction between self-esteem and self-concept. The term, self-esteem, could be limited to the evaluative aspects of the self, and the term, self-concept, could then be used for all self-descriptions which do not necessarily involve judgements of worth.
Shavelson, Hubner， οnd Stanton
In the last decade, self-concept has been recognized as a multidimensional construct. Self-concept consists of various specific dimensions of the self that reflect an individual's multipIe roles and experiences (M訂曲， 1990a)
The flrst multidimensional model of self-concept was presented in 1976 by Shavelson et al. This model has become the foundation for the study of self-concept. Shavelson et al. reported that self-concept was not an entity within the person, but a hypothetical construct that was potentially useful in explaining and predicting how a person acted. These self-perceptions influenced the way s/he acted, and these acts in tum influenced the person's self-perception. Their deflnition provided a clear and concise description of self-concept. The most important contribution by Shavelson et al. was to set up an outline of the essential features of self-concept (:ijoan, 1998), which they considered as important components in understanding self-concept.
Seven Features of Self-Concept
Shavelson et al. (1976) identified seven features that were critical to their definition of the self-concept construct:
1. Self-concept is organized or structured. This means that an individual categorizes his/her experiences and relates these categories to one another.
2. Self-concept is a multifaceted construct. The facets of self-concept are representative of the category system adopted by the individual or by a particular group of individuals.
3. Self-concept is hierarchical. This hierarchy is structured so that the base is a ve可 specific self-concept while the general self-concept is at the top. The general self-concept consists of two divisions, the academic self-concept and the nonacademic self-concept. The academic self-concept
108 . 教育心理學報
can be split into specific subject areas and then further into areas within a subject. The nonacademic self-concept can be broken down into social and physical self-concepts, which can be divided into specific 訂閱
4. The characterization of the general self-concept can be viewed as a stable construct. As one descends the hierarchy, self-concept became increasingly situation specific and, as a consequence, less stable.
5. The construct is developmental. An individual's self-conc巴ptbecomes more differen1Ìated as s/he increases in age.
6. Self-concept is evaluative. That is, it has both a descriptive and an evaluative aspect. An individual does not just describe hirr自erself but, also, makes an evaluation of himlherself in a particular situation.
7. Self-concept is differentiable from other constructs. For examp胎， academic achievement should be more highly correlated with academic self-concept th個 with social or physical self-concept.
The Structure of Self-Concept
Shavelson et al. (1976) proposed a general self-concept at the apex of the hierarchy that was divided into academic and nonacademic self-concepts. Academic self-concept was further divided into subject-specific facets of self (e.g., English and mathematics); nonacademic self-concept was divided into social, emotional, and physical self-concepts that were further divided into more specific components (e.g., physical self-concept into physical ability and physical appearance) When Shavelson et al. proposed their model, there was little empirical support for the multidimensionality of self-concept, but subsequent empirical research did support it (Dusek & Flaherty, 1981; Marsh & Shavelson, 1985). In reviews of these research, Mar吭， Byrn巴， and Shavelson (1988) conc1uded that self-concept could not be adequately understood if this mu 1tidimensionality was ignored.
Marsh and Shavelson (1985) tested the original Shavelson model with responses by Australian elementary school children to the SDQ-I. They found the hierarchy to be more complicated than anticipated. In particul缸， verbal and math self-concepts were nearly uncorrelated with each other and did not combine with the general school self-conc巴ptto form a single, second-order academic factor. Instead, they supported the division of the general self-concept into a nonacademic self-concept, an academic English self-concept, and an academic mathematics self-concept. This revision of the original model was called the Marsh-Shavelson revision. Marsh, et al. (1988) further tested their revised model and found that the pattern of correlations among the factors was consistent across responses to different self-concept instruments and supported the Marsh-Shavelson revision
Relationships Between Self-Concepts and Esteem
Some studies have revealed positive relationships between specific elements of self-concept and global self-esteem. Marsh (1986) found that students' specific self-concepts correlated at better than .69 w他 global self-的tee肌 Thefindings demonstrated that a su 句 ect' sself-concept in a
specific area was moderately correlated with his or her perceived importance of the 訂閱，and also strong tendency with his/her high self-esteem. Rosenberg (1979) reported similar finding that individual with high self-concept in particular facets would attach higher importance to these facets and that this mechanism would allow individuals to enhance or protect their esteem. That is, when the level of specific self-concept was more positiv巴， the positive contribution to global self-esteem was larger.
The sample for this study was composed of 591 junior and senior high school students in Taipei, Taiwan; there were 316 males and 275 females. At 由etime of the study, p訂ticipants were enrolled in the ninth grade, age 15 (i.e門 equivalentto third year students in junior high school in Taiwan); tenth grade, age 16 (i.e. ，巴quivalentto first ye訂 studentsin senior high school in Taiwan); . and eleventh grade, age 17 (i.e., equivalent to second year students in senior high school in T位wan).
The students who attended two junior high schools (i.e., Chu-Lin and Chung-Ho Junior High Schools), and six senior high schools (i.e., Taipei First Girls Senior High School, Jan-Chung
Senior High School and Appendant Senior High School of Normal Universi句， Hua司Chiang Senior High School, Hua-Chiao Senior High School, and Tai-Shan Senior High School) were selected as the target population.
Stratified sampling was employed to select 80-100 students (i.e., approximately 2 or 3 5巴ctions of classes) from each of the 6 senior high schools. Questionnaire were adrninistered by 650 students. However, only 591 participants completed both of the questionnaires (i.e., the Chinese Rosenberg Self-Esteem Scale and the Chinese Self-Description Questionnaire-II) and provided useful data.
Of the total sample (N
=591) who retumed the questionnaires, 53.74% (n
=316) were males, and 4的 .26% (n
=275) were females. For grade level, 36.56% of the students were in the ninth grade, 29.93% in the tenth grade, and 33.50% in the eleventh grade.
Students participated voluntarily in this research, and no reward was offered for participation, nor was there any penalty for not participating. Confidentiality was ensured by having all participants complete the questionnaires anonymously.
The instruments that were used in this study included a Demographic Survey, the Chinese Rosenberg Self-Esteem Scale (CRSES), and the Chinese Self-Description Questionnaire-Il (CSDQ-II). Originally, the CRSES was constructed by Rosenberg (1965) as an instrument to measure adolescents' global self-esteem, and the CSDQ-II was constructed by Marsh (l990b) to
measure multifaceted self-concepts of adolescents. These two inventories were translated into Chinese by Lin (1990) and Hau (1996), respectively. Exploratory factor analyses were used to access the validity for the CRSES and CSDQ-II by the researcher. Two factors, positive and negative items, w巴re conducted in the CRSES. Nine factors were extracted in the CSDQ司11 ，they were: Math/School factor, Verbal factor, Physical Appearance factor, Peer Relations factor, Emotion factor, Physical Ability factor, Parent Relations factor, Negative General Self factor, and Honest factor.
Two questionnair間，the CRSES (Lin, 1990) and Chinese SDQ-II (H仙， 1996), as well as the Demographic Survey were administered in group (i 巴，class) format. The teacher of each class or this researcher explained the pu叩oseof the study and answered any questions from students. All participants were informed of their rights of confidentiality before they responded to the questionnaires. The teachers encouraged the students to answer every question carefully and informed them that a11 responses would be anonymous; names were not writtt>n on the materials.
There was no time limit for students to answer the questionnaires; it took approximately 25 minutes for them to complete all questions. Finally, the participants were reminded to check whether they had answered every question.
Data Designs and Analyses
The data from the collected questionnaires were held anonymous and carefully examined to determine whether the completed response forms were usable for scoring and verification. The data obtained through standard measuring were treated sequentially as follows.
Correlations were used to examine the relationships between adolescents' global self-esteem, including positive and negative components, and the nine factors of domain-specific self國 concepts; the multiple regression analysis was conducted to explore what dimensions of self-concepts could predict adolescents' global self-esteem. In addition, grade and gender differences in this model were conducted separately; nine 2 (gender) x 3 (grade) ana1yses of variance (ANOVAS) designs were conducted to explore gender and grade differences in the specific domains of self-concept. A post hoc comparison was employed to examine the differences in the specific domains of self-concept among the grades.
The Relαtionships Between Domain-Specific Self-Concepts and Global Self-Esteem
The correlation coefficient was utilized to examine the relationships between the adolescents' perceived domain-specific self-concepts and global self-esteem . The appearance of the correlation matrix revealed that the variables of the tota1 CRSES scores, the positive CRSES scores, and the
臺灣青少年多重向度自我概念與整體自尊關係之比較研究 . l l I
-negative CRSES scores were significantly related positively with a11 nine SDQ-II factors and the total CSDQ-II scores.
In Table 1, the values of the correlation coefficient of the total CRSES ，出e positive CRSES, and the negative CRSES with the factor of the Negative Genera1 Self were ve可 highlycorrelated (r
=.駒， .72, and .73). AIso, the three CRSES scores provided a moderately high correlation with the remaining eight factors of the II and high correlation with the total scores of the CSDQ-II (r
=.76), but there was a weak correlation with the factor of Honesty (r
=.21, .19, and .18). In addition, the MathlSchool factor and the Physical Appe訂ance factor showed higher correlations with the CRSES factors than the other factors, which indicated that the MathlSchool factor and the
Table 1 Correlation Coefficients of the Nine CSDQ-II Factors and the CRSES Scores
Factor Total Positive Negative
The Math/School Factor .51 ** .48** 46**
The Yerbal Factor 43** .46** 32**
The Physical Appearance Factor 49** 46** .43**
The Peer Relations Factor 45** .43** 39**
The Emotion Factor 45** 37** 46**
The Parent Relations Factor 36** 32** 33**
The Physical Ability Factor .28** .31 ** .20**
The Negative General SelfFactor .80** .72** 73**
The Honesty Factor 之 1** 19** 。 18**
The Total CSDQ-II Scores .76** 73** .66**
**p < .001
Physical Appearance factor were highly correlated to the adolescents' global self-esteem. However, the factor of Emotion correlated higher in the negative CRSES (r
=.46) than in the positive CRSES (r
Three simultaneous multiple regression analyses (Cohen & Cohen, 1978) were concillClèd to examine the unique contribution of the domain specific self-concepts in the prediction of global self-esteem with alI adolescents and by gender and grade differences in the current study (Research Question 1). The scores on the nine factors of the CSDQ-II served as 也eindependent variables, while the scores on the CRSES served as the dependent variables
It was supposed that the factors in the CSDQ-II were independent from each other after an orthogonal rotation in the factor analysis. However, as shown in Table 2, the zero-order correlations (r
=.26 - .50,) showed that the factor of Negative General Self was correlated higher with the remaining eight factors than the other intercorrelations between factors in this scale,
which indicated the Negative General Self scale could be applied to each specific facet of the self In order to reduce multicollinearity、 the Negative General Self factor was eliminated in the model of the multiple regression analysis in this study
112 . 教育心理學報
Table 2 Intercorrelations Between Nine Factors ofthe CSDQ-II
Factor II III IV V VI VII VIII IX
1 Math/School 23** 27** .28** .23** .24** .28** .50** .13** II Verbal .24** .22** .12** .14** 14** 40** .11 ** III Appearance .53** 28** .30** 19** 46** 06 IV Peer Relation 38** .26** 24** 45** 21** V Emotion .23** 26** .50** .10* VI Parent Relations 08* 。 26** .05 VlI Physical Ability .38** .39**
VIII Negative General Self 20**
Note Correlations based on N = 591.
As displayed in Table 3, Factors 1 (Math/School), 11 (Verbal), III (Physical Appearance), IV (Peer Relations), V (Emotion), and VII (Parent Relations) of the CSDQ-II were the six predictors that made a significant contribution to the global se1f-esteem (CRSES). According to the standardized Beta weights of these eight predictors in Table 3, the MathJSchool factor explained
most of the variance, followed by the Verbal factor, the Emotion factor, the Physical Appearance
factor, the Parent Relations factor, and the Peer Relations factor.
Table 3 Summary of the Multiple Regression Analysis of the CRSES on the Subscales of the CSDQ-II for AII Adolescents (N = 591)
Predictor Beta Math/School .28 Verbal .25 Physical Appearance 22 Peer Relations .07 Emotio .2 Physical Ability 03 Parent Relations 10 Honesty .05 9.18 8.46 6.52 2.0。 7.47 0.97 3.22 1.64 IJ3)504ll p 一帥帥帥 MW ∞拈∞的 Note Overall F 俏， 590)= 89.67, p < .001; R2 = .55
Gender and Grade Differences in the Prediction 01 Global Self-Esteem
All the factors with the exception of Peer Relations and Physical Ability made a significant
contribution in the prediction of ninth grade adolescents' global self-esteem at 出e significance
level of.的. The standardized Beta values indicated that the MathJSchool factor appeared to be the
臺灣青少年多重向度自我概念與整體自尊關係之比較研究 113 .
Parent Relations, and Hon巴sty.
Factor 1 (Math/School), Factor II (Verbal), Factor III (Physical Appearance), Factor 1V (Peer Relations), and Factor V (Emotion) of the CSDQ-II were the five predictors at the significant level 00 global self-esteem (i.e., the CRSES) for the tenth grade students. An examination of the standardized Beta values revealed that the MathlSchool factor explained most of the variance. followed by the factors of Verbal, Physical Appearance, Peer Relations, and Emotion.
Five self-concept factors predicted the eleventh grade students' global self-esteem on the CRSES. They were Factor 1 (Math/School), Factor II (Verbal), Factor III (Physical Appearance), Factor IV (Peer Relation), and Factor V (Emotion) of the CSDQ-ll. According to the standardized Beta values, the Verbal factor explained most of the variance, followed by the Emotion factor, the Math/School factor, Physical Appearance, and the Peer Relations factor. The Honesty factor was the least influential variable that entered into the equation
Factor 1 (Math/School), Factor II (Verbal), Factor III (Physical Appearance), and Factor V (Emotion) were the four predictors at the significance level of .001 on global self-esteem (i .e. ，出E CRSES) for male adolescents. The standardized Beta weights suggested that the Math/School factor was the best predictor, followed by the Verbal factor, the Emotion factor, and the Physical Appearance factor.
1t appears that Factor 1 (MathlSchool), Factor II (Verbal), Factor III (Physical Appe缸ance)， Factor V (Emotion), and Factor VII (Parent Relations) of the CSDQ-II entered the regression equation for the female adolescents. According to the standardized Beta weights, the Verbal factor explained most of the varianc巴， fOllowed by the Emotion factor, the MathlSchool factor, Physical Appearance factor, and the factor of Parent Relations. The Physical Ability factor was the least influential variable that entered into 伽 equation.
Gender and Grade Differences in the Domain-Specific Self-Concepts
Nine 2 (gender) x 3 (grade) analyses of variance (ANOVAS) were performed to analyze the adolescents' perception of their functioning on each factor of the CSDQ-II separately (Research Question 2 and 3). Significant main gender effects were found for Math/甸hoolfactor [F(2, 590)
12.33, P < .001], Emotion factor [F(2, 590)
=5.62, p < .0旬，Physical Ability factor [F(2, 590)
=11.04, P < .001], and Honesty factor [F(2, 590) = 9.28, p < .05] of the CSDQ-11.
Presented in Table 4, exarnination of the means showed that boys had significantly greater confidence in the Math/School factor (M
=73.34) than girls (M
=67.91) and perceived themselves to be more stable in the Emotion factor (M
=35.85) 也andid girls (M
=34.20). The boys' scores on the Physical Ability factor (M
=30.09) were higher than the girls (M
=28.15), but the girls had significantly higher Honesty scores (M
=37.98) than the boys (M
=36.16). There were no significant differences between boys and girls in the self-concept factors of Verbal, Physical Appearance, Parent Relations, and Negative General Self.
報 學 理 'L‘ 育 教 114
Means and Standard Deviation for Nine Factors of the CSDQ-II by Gender Table 4 Girls (n = 275) Boys (n = 316) M Self-Concept SD M SD Factors 19.72 67.91 20.07 73.35 MathlSchool 10.15 10.71 39.90 42.21 10.59 10.63 40.72 42.54 Verbal PhysicaI Appearance Peer Relations 3.83 8.41 53.99 9.12 9.56 8.28 34.20 28.15 9.58 .22 35.85 30.09 Emotion PhysicaI Ability Parent Relations 34.35 .69 34.70 .79 5.47 21.99 5.50 22.50
Negative GeneraI Self
In additi凹， there was a significant interaction e釘ect between grade and gender for the Peer Relations factor [F(2, 590) =4.91, p < .05} (see Figure 1). The boys in the tenth grade scored higher than males in the other grades, and the girls in the ninth grade scored higher than females in other grades.
Overall, the findings of the contribution of the CSDQ-II on the adolescents' global
self-esteem are listed as follows.
37.98 7.10 36.16 girl boy O 口 56 54 52 曲曲』 oum 11 nu 1 9 Grade
Figure 1. The Interaction Effect on the Factor of Peer Relations Between Gender and Grade
1. Academic factors (Math/School and Verbal), the Emotion factor and the Physical
Appearance factor contributed the most variance to the CSDQ-II for gender and rade.
2. The factors of Parent Relations and Honesty provlded less contribution to the global self-esteem in increasing grade. In other words, Parent Relations and Honesty affected adolescents' perception of their global self-esteem only in the ninth grade, but not in the tel!th and eleventh grade.
3. Adolescents' perceptions of their global self-esteem were affected by peer relations in the tenth and eleventh grade but not in the ninth and tenth grade, which indicated that peer relations
became more important than parent relations for adolescents as they grew older.
4. The factor of Parent Relations predicted a higher correlation in global self-esteem with girls than with boys. Female adolescents had a higher connection wíth theír parents than male adolescents.
The Relatíonships Between Domain-Specific Self-Concept and Global Self-Esteem
Little empirical research has been conducted to examine how different domains of self-concept differentially contribute to global self-esteem (Marsh, 1986). Rosenberg, Schooler, Schoenbach, and Rosenberg (1995) reported that academic self-concept, positi ve family relationships, the number of best friends, and physical appe訂ance were key variables that directly affect global self-esteem. This current study, which was conducted with 591 subjects in Taiwan, demonstrated that six domain-specific self-concepts (i.e., the factors of MathJSchool, Emotion, Physical Appearance, Verbal, Parent Relations, and Peer Relations) in the CSDQ-II greatly contributed to the prediction of global self-esteem; in partícular, the factor of Math/School contributed most. These results were similar to Westem findings (Marsh, 1990b; Mar曲， et al., 1988; Marsh , Parker, & Bam凹， 1985; Marsh & Shavelson, 1985; Rosenberg et al., 1995), and other descriptíons of the relationship between these six domains of concept and global self-esteem are stated as follows.
Math/School and 始 rbal
In the correIation ma甘睬，出efactor of MathJSchool correlated highly with global self-esteem, wíth th巴 positiveand negative components, and with the total scores, as well as the factor of Verbal in the second highest correlation with the CRSES (Lin, 1990). The results ín the muItiple
時remonanalysisindmted that Math/School and Verbal self< oncfpts
Cωont甘盯r口i胎bu叫I此tlωon肘1芯s tωo pr昀ed副icti昀onof global seIf-esteem. Also, researchers in the Uníted States have found that academic performance was positively related to global self-esteem (Rosenberg, Schooler, & Schoenbach, 1989; Wyl時， 1979), as well as the positíve connection between academic self-concept and global self-esteem (Marsh, Byme, & Shavelson, 1988; Marsh et al.、 1985; Marsh & Shavelson, 1985).
GeneraJly, education is a major area of achievement among adolescents. School performance is a particularly visible indicator of a student's worth and capability; and academic success is valued in socie句， especially in Chinese culture. As mentíoned in the Iiterature review, high standards of educatíon were the most important values of students in Hong Kong and Taiwan (Cheng, 1995; Kong, Hau. & Cheng, 1998; Lin, 1997: Tumer & Mo, 1984;); therefore, it was not su叩risingthat academic seIf-concepts in the Chinese SDQ-II (Hau, 1996) were strongly correlated
116 . 教育心理學報
with global self-esteem. This result was consistent with the Rosenberg et al. (1995) finding that academic self-concept had a more powe的11effect on global self-esteem 血anthe cther variables
For gender differences，也e current findings supported Stevenson and Newman' s (1986) findings that boys scored higher in the factor of MathJSchool than girls. In Taiw切， the education of all elementary and junior high sch∞1 age chil的n is compulso旬， which means a free public education is provided for the first through the ninth grades. Therefore, it is only after this compulsory education that p訂閱ts have to make decisions about the cost of education and malce choices about whether ~ son or daughter is provided with further education. The expected gender preferences dìfference may show up at a later age. If the son faìls to pass the high school entrance examination, p紅ents may give him preference by spendìng whatever it takes for him to attend a cram school in order to retake the examinations in the following year. Howev仗， if the daughter ìn the same family faìls, sh巴的 morelikely to be channeled into a vocational school. Of course, this is dependent upon the funds avaìlable to 也efamily.
Moreover, the main effect for grade was conducted in the factor of Math/School in the current study. The tenth grade adolescents scored higher on this factor than the other two grades. As mentioned earli仗， the competitive entrance examination causes extreme pressure for junior and senior high school students in Taìwan. Since the ninth grade students in this study wiU take the entrance examination for senior high school ìn the following summer, most of them were concerned about their academic abilities. Also，也eyworried about having to go to the cram school, which is expensive, if they failed 也isexamination.
In contrast to 由eninth grade students, the tenth grade students felt relieved because they had passed 也eexarnination and entered one of the public senior high schools the previous ye也Thus ， these tenth grade adolescents were confident and received high scores in most of the domain-specific self-concepts in the CSDQ-II.
For the eleventh grade students, who were close to the time for taking the entrance examination for colleges, this event represents another competition experience, which caused them more pressure and anxiety than the tenth grade adolescents. Hence, more stress and anxiety were reported by the eleventh grade students. These adolescen
An examination of the correlation coefficien郎，pertaìning to the relationship between emotion and global self-esteem, showed a significant relationship to 巴achother (Rosenberg et a1., 1989; 缸， 1980). Negative globa1 self-esteem correlated highly with emotion in this study, whìch was sìmilar to the findings in Fleming and Courtney's (1984) study that self-esteem negatively correlated wìth anxiety and depression. On the other hand，的 withsimilar research (Covey & Feltz, 1991; Hsieh, 1999), boys scored higher emotiona1 stability 也angirls in the current study.
Some researchers have noted that boys' stability in emotion might be affected by a function of the stereotyped masculine gender role that prescribes the avoidance of self-disc1osure in boys. Zeman and Shipman (1997) pointed out that boys reported that they dissembled in regard to
臺灣青少年多重向度自我概念與整體自尊關係之比較研究 117 .
ernotion rnore than girls. This finding was consistent with gender role stereotypes, which irnplies that rnasculinity is reduceó wh巴n boys talk about their feelings. In contrast, the stereotyped ferninine ernotional display rule encourages girls to engage in girl talk (PoIce-Lynch, Myers, & Kilrnartin, 1998). Fernales who express ernotions rnay receive rnore support frorn others because social support tends to assist fernales. It rnay be girls are less restrictive in terrns of ernotional expression becaus巴 they have a greater facility with verbal tasks (e.g., reporting thoughts and feelings are easier for thern).
The younger adolescents (i.e., ninth grade) in Taiwan had a higher stable emotion than the older adolescents in this study. One explanation for the age related decrease in emotional self-concept could be the influence of environmental factors. It appeared that students' emotional selι concept decIined during the period of senior high school. Because of the different tracks for senior high schools, adolescents rnay experience rnore pressure than they did in junior high school. School labeling at the senior high school level caused adolescents to experience more pressure than in junior high school. If students enter the high track school, their p訂閱的， teachers, and
themselves establish a higher goal to pursue adrnission to the university.
Harter (1999) noted that physical appe缸個ce correlated rnost highly with global self-worth (i.e., frorn .52-.80), but this finding was not consistent with the results from this current study because, here, the factor of physical appe訂ance was correlated as the fourth highest with globa1 self-esteern. However, cultural differences rnay explain the disparity in the results. Crystal, Watanabe, Weinfu此，and Wu (1998) revealed that a higher proportion of U.S. adolescents reported
human differences in evaluation dornains, such as appearance and attractiveness; however, most of the Asian adolescents, in their study, valued high acadernic achievement and cognitive ability.
From this comparison of these two cultures (Crystal et al., 1998), it appe紅sthat rnembers of the Chinese culture emphasize greatly the importance of acadernic achievernent, which has a positive relationships with personal success, and the adolescents in Taiwan emphasize acadernic performances more than adolescents in other cultures. In other words, successful academic achievernent will supplement the inferiority of physical appe紅anceand wðl bring benefits in the future. Also, it was demonstrated in the interview results 也atnone of the respondents reported that they liked themselves because of physical appearance.
Sorne researchers have found 曲的 relationshipswith peers (Bemdt & Keefe, 1998; Cheng & Chang, 1994; Chu, 1981; MccIun & Merrell, 1998) were highly correlated to global self-esteem
Bemdt and Keefe revealed that students, who described their friendships as having more positive features, perceived their global self-worth more positively during adolescence. Specifically, in this study, the older adolescents reported a higher correlation wíth peers than did the younger groups, sirnilar to the results found by Marsh, Smith, Marsh, and Owens (1988) and Hattie (1992)
In addition‘ Chinese adolescents are not encouraged to make friends with opposite sex peers
118 . 教育心理學報
less attention to their academic performanc巴 ifthey date friends of the opposite sex. That may be the reason that there was no distinction in the peer relations in this study.
There were no gender and grade main effects in the factor of peer relations, but an interaction effect by gender and grade was found in this study. The girls in the ninth grade and the boys in the tenth grade scored highly on the factor of peer relations. This may be due to the fact that, in the ninth grade, girls have had ongoing, intimate relationships with their c1assmates for 3 ye紅s. When girls enter senior high school, they have to make new friends and build new relationships based on environmental differenc~s; however, after th~y adjust to the new e盯ironme此， the development of
peer relationships increases quickly. In contrast, the intimacy of boys' peer relations is more related to gender role identity, that 蹈， boys who identify strongly with the traditional masculine
role are less likely to develop strong peer relations (Jones & Dembo, 1989).
A nl1mber of researchers have demonstrated that the relationships with parents and global
self-esteem were highly correlated (Ho, 1992; Huang, 1997; Ju, 1996; Mcc1un & Merrell, 1998; Wu, 1998). Adolescents who perceived their parents as having an authoritative p缸entingstyle had a more positive self-concept than those who perceived their parents as having an authoritarian orientation (Mcc1un & Merrell). Although parenting styles were not addressed in the present study, other findings for Taiwanese students showed a similarity with Western results. The factor of parent relations was highly correlated with global self-esteem (Marsh et al., 1988), and younger adolescents scored higher in the factor of p紅entrelations than did the older ones (Hatt時， 1992).
For most of the CSDQ scales, gender effects were statistically significant; some favored girls, but more favored boys, the same as in Marsh's (1989) study. As 巴xpected， a significant g巴nder effect on the specific domains of self-concepts was found in this study, which indicat巴d that boys had higher scores in most of the domain-specific self-concepts (i.e叮 the factors of Math/Sc11001, Emotion, and Physical Ability) than girls. Girls scored significantly higher th祖 boys only for the factor of Honesty. The same results were found in the research conducted in Australia (Marsh, 1990b) and Shanghai (Cheng, Zhu, Ye, & Tang, 1997). Conclusive旬， the direction of gender difference in specific domains of self-concept tended to be consistent with traditìonal gender stereotypes (Dusek & Fl油erty， 1981; Wylie, 1979) and, in spite of cultural differences, boys had higher self-concepts in masculinity and achievement than girls, but lower self-concepts in congeniality.
RECOMMENDATIONS FOR FUTURE RESEARCH
Although the Chinese SDQ-II (Hau, 1996) may have relatively high validity and reliability in the measurement of multidimensional self-concepts in these adolescents in Taiwan, research with more participants from different areas in Taiwan is needed to further determine the validity and reliability and establish norms for this instrument. Many rese紅ch studies have been conducted in
出earea of self-esteem, but few researchers have repo此edspecific definitions of high and low self-esteem. Participants' responses about high and low self-esteem were self-reports; therefore, the findings may be ambiguous.
Bemdt, T. J., & Keefe, K. (1998). Effects of friendship on adolescents' self-esteem. Purdue University, Department of Education. (ERIC Document Reproduction Service No. ED 431514)
Bo阻， C. H. (1998). An investigation of differences in global self-concept and specific domains of self-concept based on age group, gender, and gender role η'pe. Unpublished doctoral dissertation, University of Alabama, Tuscaloosa.
Brandon, N. (1969). 了于ze psychology of self-esteem: A new concept of man 's psychology. Los Angeles, CA: Nash.
Byrne, B. M. (1996). Measuring self-concept across the life span issues and instruction. Washington, DC: American Psychological Association.
Ch妞， D. W. (1997). Self-concept domains and global self-worth among Chinese adolescents in Hong Kong. Personaliη and lndividual Differences, 22, 511-520.
Ch阻， D. W.,
&Lee, H. C. B. (1993). Dimensions of self-esteem and psychological symptoms among Chinese adolescents in Hong Kong. Joumal of Youth and Adolescence, 22, 425-440.
Cheng, L (1995). The relationships among vocational high school students and anxiety, self esteem and locus of control. Student Joumal ofTainan College， 蹈， 22-38. [in Chinese]
Cheng, L. Z., & Chang, Y. L. (1994). The relationships among children-parent, peers and self-concept in the 5th and 6th grade students. Joumal of Tainan 1切cher
sCollege， 刀， 1-18. [in Chinese]
Cheng, g., Zhu, X., Ye, L., & Tang, Y. (1997). The revision of the Shanghai norm of the
-Description Questionnaire. Psychological Science (China), 20, 499-503.
Chu, G. M. (1981). The relationship between self-concept and peer relations injunior high school students. Unpublished master's thesis, National Taiwan Normal Universi句，Taipei.
Cohen, P., & Cohen, E. (1978). Selecting an Appreciate Model for Data Analysis. Mult伊leLiner Regression Viewpoints, 8, 3, 106-115. (ERIC Document Reproduction Service No. EJ
Cooley, C. H. (1902). Human nature and the social order. New York: Scribner.
Coopersmith, S. (1967). The antecedents of self-esteem. San Francisco, CA: W. H. Freeman. Covey, L. A., & Feltz, D. C. (1991). Physical activity and adolescent female psychological
development. Joumal of Youth and Adolescence, 20, 463-474.
Crystal, D. S 門 Watanab巴，日， Wein如此， K., & Wu, C. (1998). Concepts of human differences: A comparison of Americ姐， Japanese, and Chinese children and adolescents. Developmental
120 . 教育心理學報
Psychology, 34, 714-722.
Daly, 1. A., & Diesel, C. A. (1992). Measures of comrnunication related personality variables. Communication Education, 41,405-414.
Dusek, J. B., & FI祉lerty， J. F. (1981). The development of self-concept during adolescent years. Monographs ofthe Society for Research in Child Development, 46(4, Serial No. 的 1) Elliott, G. (1986). Self-esteem and self-presentation among the young as a function of age and
gender. Joumal of 跆uthand Adolescence， 刃， 135-153.
Fleming, 1. S., & Codrtney, B. E. (1984). The dimensionality of self-esteem: Hierarchical facet model for revised measurement scales. Joumal of Personality and Social psychology, 46,
Grant, T. (1998). A consideration of select dimensions of self-concept and their relation to global self-esteem in native-bom African-American and Caucasian young (adolescents, inner-ciη1). Unpublished doctoral dissertation, New York University.
Hamid, P. N., & Cheng, C. (1995). Self-esteem, and self-concept clarity in Chinese students. Social Behavior and Personality, 23, 273-284.
Harter, S. (1999). The constructiθn ofthe se
lf.New York: Guilford Press. Hatti巴，J. (1992). Self-concept. Hillsdale, NJ: Erlbaum.
Hau, K. T. (1996). Chinese students' self-concept: Factor structure, stability, and factorial invariance across grades. Hong Kong: The Chinese University of Hong Kong, Department of Education.
Ho, Y. (1992). The effect of child's self-esteem and misbehavior from parent's disciple in d~加rent family structure. Unpublished rnaster's 由esis，Wen-Hua University, Taipei, Taiwan. Hoelter, J. W. (1985). The structure of self-conception: conceptualization and measurement.
Joumal of Personality and Social Psychology, 47, 2 日司 23 l.
Hsieh, H. D. (1999). Self-concept, emotional stability and learning of students in colleges. Joumal of Life and Science, 5, 1-22. [in Chinese]
Huang, L. J. (1971). Sex role stereotypes and self-concepts among Chinese and American students. Joumal ofComparative Family Studies, 2(2),215-234.
Huang, T. (1997). The relations among parent-discipline, self-concept, behavior delinquency for junior high school students. Digest of Education Information, 40(3),22-33. [in Chinese] Iheanacho, S. O. (1988). Minority self-concept: a research review. Journal of Instructional
James, W. (1890). Psychology. New York: Fawcett.
臺灣青少年多重向度自我概念與整體自尊關係之比較研究 121 .
Lau, S., & Cheung, P. C. (1987). Relations between Chinese adolescents' perception of parental control and organization and their perception of parental warmth. Developmental Psychology, 23,726-729.
Lin, R. (1990). Reliability and validity of the Rosenberg Self-Esteem Scale on Chinese children. Joumal of National Chung-Chi University: Volume of Social Science, 29-46. [in Chinese] Lin, Y. (1997). Academic achievement and self-concept: Samples of senior high school students in
Japan and Taiwan. Unpublished master's thesis. Kaohsiung Medical College, Kaohsiung, T也 wan.
Marsh, H. W. (1986). Global self-esteem: Its relation to specific facets of self-concept and their importance. Joumal of Personality and Social Psychology， 刃， 1224-1236.
Marsh, H. W. (1989). Age and sex effects in multidimensional self-concepts: Preadolescence to early adulthood. Joumal of Educational Psychology， 肘， 417-430.
Marsh, H. W. (l990a). The structure of academic self-concept: The Marsh/Shavelson model. Joumal of Educational Psychology, 82,623-636.
M紅油，H. W. (l990b). Self-Description Questionnaire (SDQ-II): A theoretical and empirical basis for the measurement of multiple dimensions of adolescent self-concept. An interim test manual and research monograph. Macarthl旺， New South Wales, Australia: University of
Westem Sydney, Faculty of Education.
Marsh, H. w., Byme, B. M., & Shavelson, R. J. (1988). A multifaceted academic self-concept: Its hierarchical structure and its relation to academic achievement. Joumal of Educational Psychology, 80, 366-380.
Marsh, H. W.叮 Parker， J., & Bames, 1. (1985). Multidimensional adolescent self-concepts: Their relationship to age, sex and academic measures. American Educational Research Council,
Marsh, H. w., Smith, 1.口， M訂曲， M. 氏， & Owens, L. (1988). The transition from single-sex to coeducat.ional high schools: Effects on multiple dimensions of self-concept and on academic achievement. American Educational Research Joumal, 25,237-269.
M紅油， H. w., & Shavelson, R. J. (1985). Self司concept: Its multifaceted, hierarchical structure. e
Educational Psychologist, 20, 107-125.
Mcc1un, L. A., & Merrell, K. W. (1998). Relationship of perceived parenting styles, locus of control orientation, and self-concept among junior high age students. Psychology in the Schools, 35, 381-390.
122 . 教育心理學報
Rogers, C. (1951). Client-centered therapy: Its practice, implications and theory. Boston: Houghton Mifflin.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton, NJ: Princeton University Press
Rosenberg, M. (1979). Conceiving the self. New York: Basíc Books.
Rosenberg, M., Schooler, c., & Schoenbach, C. (1989). Self-esteem and adolescent problems: Modeling reciprocal effects. American Sociological Review, 54, 1004-1018.
Rosenberg, M., Schooler, c., Schoenbach, c., & Rosenberg, F. (1995). Global self-esteem and specific self司 esteem: Different concepts, different outcomes. American Sociological Review, 60, 141-156.
Shavelson, R. J., Hubner, J. J., & Stanton, G. C. (1976). Self-concept: Validation of construct interpretation. Review of Educational Research， 咐， 407-44 1.
Stevenson, H. W., & Newman, R. S. (1986). Long-term prediction of achievement and attitudes in mathematics and readíng. Child Development, 57, 649-659.
缸， c. W. (1980). Emotional development of children and adolescents. Joumal of Educational Psychology, 12,99-113. [in Chinese]
Tumer, S. M., & Mo, L. (1984). Chinese adolescents' self-concept as measured by the Offer-Self-Image Questionnaire. Joumal of Youth and Adolescence, 13(2), l31-143.
Watkir芯，旦， & Dhawan, N. (1989). Do we need to distinguish the cons甘ucts of self-concept and self-esteem? Joumal of Socíal Behavior and Personality, 4, 555-562.
Wu，可(. S. (1998). The verification of the self-esteem model of adolescents. Unpublished master's thesis, National Taiwan Normal Universi句，Taipei.
Wyl時，R. C. (1979). The self-concept (Vol. 2). Lincoln: University of Nebraska Press.
Yang, K. S. (1981). Social orientation and modemity arnong Chinese students in Taiwan. Joumal ofSocial Psychology, Jl3, 159-l70.
Zeman, 1., & Shipman, K. (1997). Social contextual influences on expectancies form managing anger and sadness: The transition from rniddle childhood to adolescence. Developmental Psychology, 33, 9l7-924
收稿日期: 2000 年 9 月 18 日 接受刊登日期: 2∞ 1 年 6 月 29 日
Bulletin of Educational Psychology, 2∞2 ， 33(2) ， 103-124
National Taiwan Normal University, Taipei, Taiwan, R.O.C
整體自尊關係之比較研究朱玉去尾 中團技術學院 企業管理系 摘要 123 本研究主要目的在探討青少年整體自尊和多重向度自我概念之關聯性。作者使用中文版的「自我 概念量表 J (Self-Description Questionn位re-II)及「羅氏自重量表 J (Rosenberg Self-Esteem Scale)為測量 工具，收集台灣地區 591 名園、高中學生的作答結果，並採用相關( correlation) 、多變項迴歸分析
(multiple regression analysis) 、及變異數分析 (ANOVA) 等統計分析研究資料。主要的發現如下: (1)自
我概念量表各因素和羅氏自重量表的得分有顯著相關; (2) 學科/數學、語言、外表、情緒穩定性、親 子關係、和同儕關係等六項多重向度自我概念可以明顯預測整體自尊; (3) 青少年因年齡及性別之不 同，其知覺到的多重向度自我概念也有所差異; (4) 男女生因年齡不同在向儕關係自我概念中呈現顯 著的交互作用，而在其他多重向度自我概念及整體自尊上則無。 上述的研究發現有助於教育工作者、學校諮商人員、和教育研究者思考並注重各個向度的自我概 念對青少年整體自尊發展的影響。作者建議未來對可針對多重向度自我概在作進一步的研究，例如發 展適用於台灣地區青少年多重向度自我概念的測量工具，以及設計提昇青少年自尊的教育課程等。 關鐘詞:自尊、多重向度自我概念、性別、年齡、自我概念量表